If you separate fields on import do you need to render fields on export?

[Chris Wright]

You only sep. fields if it is interlaced footage, and if so, preserve best edges box is best quality.

If you sep. fields on a progressive clip, you are reducing its quality because AE is trying to sep. fields that don’t exist.

When you render frames only, you are asking AE to deinterlace all your interlaced footage. Progressive is passed along normally in this fashion though.

If you don’t want to lose quality with interlaced video, interpret it as field(dominant upper or lower) and render it out to a new field dominance(most often the same) then you can add pulldown if necessary for NTSC. WWSSW is correct telecine for film to tv, for example.

[Kevin Camp]

as chris describes, checking the option for preserve edges should help to clean up the edges of interlaced footage if you want to render as progressive. you may also be able to further smooth the footage by enabling frame blending (frame mix) which will blend in data from the other field that you are losing. however, if there is a lot of movement in the footage, this may produce a bit of a double-image-like effect (similar to that of echo or cc wide time), so it will depend on the shot.

if you want to render back to interlaced you have a few options:

1. if you are just color correcting the footage or adding static elements, you can just not separate fields and then render as normal. as long as you did nothing to effect the interlacing (like scale, reposition, rotate, blur, etc) this should work fine.
2. if you are adding moving items to the footage (maybe text flying in or something), you can separate fields (no preserve edges) then render back to fields. this will maintain the look of the footage and also generate graphic movement that is also interlaced and should better match your footage movement.
3. if you doing anything to the footage that may interfere with the integrity of the interlacing (as mentioned above), then you will want to separate fields with preserve edges on (and if you like, the frame blending option described above). at that point you can render back to fields or not, your choice.

Kevin Camp
Senior Designer
KCPQ, KMYQ & KRCW

[Terry O’brien]

Hi,

I work exclusively with interlaced footage sourced from DV, HDV or Beta. I know how to choose the correct field dominence.

The core of my question has to do with “best practices” when dealing with interlaced footage in AE. I gather from these responses that if you separate fields on input you should render to fields on output in order to maintain full resolution. Right?

By default, I have always separated fields with “preserve edges” no matter what I was doing. It seemed like the safe thing to do regardless of whatever rendering hit that it caused.

It is only recently that I noticed that my processed movies, rendered on frames, were half resolution. As Kevin mentions, if you don’t separate fields on input the resulting image rendered as frames has more detail (presuming that you haven’t scaled the source footage). If you DO separate fields on source footage, does rendering the output on fields yield the full resolution of the source clip (i.e. both fields of information)?

Obviously, I can test some scenes to confirm this, but the answer to this question seems like it should be a standard workflow issue that has already been figured out.

Thanks

[Bruce Rudolph]

One of Andrew Kramers first tutorials had an interesting alternate to the “separate fields/ interpret footage” approach.
It may yield better results by not loosing the info in the field that’s discarded.
Have a look here:
https://www.videocopilot.net/tutorials/deinterlace_in_ae/

Cheers, Bruce

[Chris Wright]

***If you DO separate fields on source footage, does rendering the output on fields yield the full resolution of the source clip***

You know, I was just about to post about deinterlacing! It seemed everyone was tip toeing around the issue of AE wrecking your video quality if you rendered frames only and it looked like that you have used this method before to get a certain look. Fortunetely, there are much better options than AE for getting that look, which have much better final image quality. Revisionfx has fieldskit to deinterlace 29.97 interlaced to progressive. Magic Bullet has one, as does Algolith. For me, particularly, going from 29i to 23.976p was tricky involving Algolith and twixtor to get 29p then interpolating to 23.976p with the right amount of blur and fluidity you expect in film.